The present invention relates to processing documents in an image-based document processing system, and is particularly directed to method of processing image data to reorient a document image represented by the image data in an image-based document processing system, such as an image-based check processing system, and an apparatus therefor.
A typical image-based check processing system includes a number of different types of workstations. The different types of workstations may include a document preparation workstation, an image capture workstation, a recognition workstation, a keying and balancing workstation, and an encoding workstation. Items such as bank checks are initially prepared at the document preparation workstation (e.g., removing staples, paper clips, and the like) for further processing at workstations downstream from the document preparation workstation. Each prepared item is fed along a document feed path of the image capture workstation which captures front and rear images of the item. The image capture workstation creates units of work and submits the created work to a workflow manager in a known way.
Workstations downstream from the image capture workstation poll the workflow manager in a known manner for work to perform, and may also create units of work which are submitted back to the workflow manager. The different workstations downstream from the image capture workstation process the front and rear images of each item. More specifically, the front and rear images of each item are sent to the recognition workstation which provides recognition results associated with the item. Operation of each workstation downstream from the recognition workstation, such as the keying and balancing workstation, and the encoding workstation, depends upon the recognition results provided by the recognition workstation.
When an image of an item is sent to the recognition workstation, the image is sometimes not oriented in a correct orientation for viewing by a human operator. This occurs because hardware associated with the image capture workstation does not scan the item in a right-side up orientation. When an image of an item is not oriented in a correct orientation, grayscale image data which is representative of the document image cannot be binarized and compressed for subsequent storage in an image archive. Eventually the misoriented image of the item needs to be oriented, either manually or automatically by a processor, before the grayscale image data which is representative of the image of the item can be binarized and compressed.
A typical misoriented image of an item needs to be rotated 90 degrees, 180 degrees, or 270 degrees to proper orientation for viewing by the human operator. To rotate the misoriented image of the item to proper orientation for viewing by the human operator, the grayscale image data which represents the image of the item needs to be processed and reoriented. The amount of grayscale image data representative of the misoriented image of the item is usually quite large. For example, a front image of a typical check is represented by about 1200×600 pixels of grayscale data. With such a large amount of grayscale image data to process, the rotation of the front image of the check may take a relatively long time. It would be desirable to process the grayscale image data in a manner which quickly reorients the image of the check to proper orientation for viewing by the human operator.